The method given at the start of the post seems to be what you need i.e .rather than reduce by a % of demand over time periods, you instead reduce the forecast by the amount of the actual orders received.
All forecasts are wrong in principle (otherwise they would be orders) the question is whether its the quantity that is wrong i.e. I know the period of the forecast is December but the quantity is just a forecast , or I know the quantity exactly - because I have agreed a contract but I don't know exactly when/at what rate the order will be called off.
The question is then when will you decide to reconcile and what will you do with the difference? some products are seasonal umbrellas/ice creams , others have more consistent demand and while there may be some stock build ups and rundowns in the supply chain week by week over a quart sales may be consistent and you may want to roll forward some unused forecast.
Forecast are often a self fulfilling prophecy anyway - if you don't forecast enough, then you will not be able to sell extra and your forecast will then be 100% , if you sell less than forecast then you may well cut costs/promote the time and reduce future forecast so over successive periods the forecast again looks accurate.
The approach to use varies by item and depends on many factors - shelf life, storage cost/space, lead time and ease of replenishment, competition, price elasticity, margins etc.
Will you focus enough to fill up the factory and then target sales and marketing to sell it? Or will you make to order and only forecast components? Are margins high enough that you cannot afford to lose a sale> is the market bug enough that you can always sell all you can make if the price is right? re=tc
That is the basis of sales and operations planning and Ax provides some useful features but like all erp system it has challenges - e,g if you forecast 1000 for the period and you take an order for 999 on the first day then is that part of the forecast or not? and who will decide? ... the order entry clerk?
If you sold milk to babies two years ago and they are now 2 years older, and competitors have entered the market then is the history projection useful?
Is it more important to forecast capacity and utilization, or components, or semi finished goods or end items- I should every item be forecasted? and with same level of precision? - does a forecast need to be 100% of expected demand or is it more prudent to forecast more to maximise sales or less to reduce risk?
The set up options needed are much clearer after such an analysis and the purpose of some of the features provided also then becomes clearer.
